The speed of propagation, which, as we have seen when discussing the measurement of the velocity of sound, depends very little on the musical note, is not at all the same in the case of the various radiations which can be propagated in the same substance.

The index of refraction varies with the duration of the period, or, if you will, with the length of wave _in vacuo_ which is proportioned to this duration, since _in vacuo_ the speed of propagation is entirely the same for all vibrations.
Cauchy was the first to propose a theory on which other attempts have been modelled; for example, the very interesting and simple one of Briot.

This last-named supposed that the luminous vibration could not perceptibly drag with it the molecular material of the medium across which it is propagated, but that matter, nevertheless, reacts on the ether with an intensity proportional to the elongation, in such a manner as tends to bring it back to its position of equilibrium.

With this simple hypothesis we can fairly well interpret the phenomena of the dispersion of light in the case of transparent substances; but far from well, as M.Carvallo has noted in some extremely careful experiments, the dispersion of the infra-red spectrum, and not at all the peculiarities presented by absorbent substances.
M.Boussinesq arrives at almost similar results, by attributing dispersion, on the other hand, to the partial dragging along of ponderable matter and to its action on the ether.

By combining, in a measure, as was subsequently done by M.Boussinesq, the two hypotheses, formulas can be established far better in accord with all the known facts.
These facts are somewhat complex.